# Deviations from the registration — PPN World Research Paper No. 3

Each change made to the registered plan or code after registration
(2026-11-11, commit 15dc3a17), numbered, dated and explained. The output of
the code as registered is kept beside the corrected one.

## Deviation 1 — 2026-11-11: robustness R6 read the wrong publisher label

`7-analyze.mjs` split releases into "distributors' websites" and
"organizations' own websites" (§8, R6) by testing Paper No. 2's publisher
type for `"distributor"`. Paper No. 2's census labels those websites
`"wire"`, so every release fell in "own websites" and the distributor half
was empty. The test now reads `"wire"`. Only R6 changes; no estimate or
hypothesis in §7 reads this label. Found on the first run of the registered
code: 743 of the 770 frame releases are on distributors' websites, 27 on
organizations' own. Registered output: `analysis-registered.json`.
